|
Briefing
|
- One Million PCM Devices – IBM Research used one million phase change memory (PCM) devices, form of non-volatile random access memory (RAM) that stores data by altering state of matter from which device is made, to run machine learning algorithm unsupervised
- In-Memory Computing - Made from germanium antimony telluride alloy, can simultaneously run computational tasks compared to standard von Neumann systems that have separate memory and processing units
- 200 Times Faster Systems – Results in systems 200 times faster and more energy efficient than advanced classical computers, applicable for machine learning computations
- Two Tasks – Algorithm successfully made temporal correlations of unknown data, such as pixel data to create drawing of British mathematician Alan Turing, as well as U.S. rainfall data covering six-month period
|
|
Accelerator
|
|
|
Sector
|
Information Technology
|
|
Organization
|
IBM Research Inc.
|
|
Source
|
|
|
Original Publication Date
|
October 24, 2017
|
